This study employed a systematic review and meta-analysis to evaluate the impact of various interventions on psychological and clinical outcomes in women experiencing pregnancy-related challenges. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed to ensure transparency and rigor in the review process. Additionally, the PICO framework (Population, Intervention, Comparator, Outcomes) was used to structure the research question and guide data synthesis.

The PRISMA flowchart mapped the study selection process as follows: •

A total of 1,200 records were identified through database searches (PubMed: 600, Embase: 400, Cochrane: 200) and an additional 40 records were retrieved manually.

The PICO framework structured the study as follows: •

Population: Women experiencing early pregnancy loss, recurrent pregnancy loss, undergoing assisted reproductive technologies (ART), or receiving antenatal care.

Comprehensive searches were conducted in PubMed, Embase, and the Cochrane Library using a combination of medical subject headings (MeSH) terms and keywords. Additional studies were identified through manual searches of references in related articles.

•

Inclusion criteria:

A standardized data extraction form was used to collect study characteristics (e.g., study design, year, location, population size), intervention details (e.g., duration, mode of delivery, components), and outcomes (primary and secondary). Two reviewers independently extracted the data to ensure accuracy and consistency.

The quality of included studies was assessed using validated tools: •

The Cochrane Risk of Bias Tool was applied to RCTs.

Quantitative data were synthesized using a random-effects meta-analysis to pool effect sizes. Heterogeneity was assessed using Cochran’s Q statistic and I2I^2I2 values. Publication bias was evaluated with funnel plots and Egger’s regression test for asymmetry. Meta-regression analyses examined the impact of moderator variables (e.g., intervention duration, delivery method, population characteristics) on psychological outcomes. Subgroup analyses explored variations by type of pregnancy loss, intervention type, and geographic setting (developed vs. developing countries).

Qualitative data were synthesized using thematic analysis to explore patient and provider experiences, as well as barriers and facilitators. Data were coded and categorized into key themes.

Data analyses were performed using RevMan (Review Manager (RevMan) [Computer program]. Version 5.4, The Cochrane Collaboration, 2020.) ¹⁷ for meta-analysis, R software (R Core Team. (2024). R: A language and environment for statistical computing (Version 4.4.2), Python (Python Software Foundation. (n.d.). Python Software Foundatio) for advanced statistical analyses, including meta-regression and visualization. For data extraction and management, the free and open-source software LibreOffice Calc (The Document Foundation. (n.d.). LibreOffice Calc (Version 7.5))was used as an alternative to proprietary spreadsheet tools. This methodologically robust approach, underpinned by the PRISMA guidelines and PICO framework, ensured comprehensive and transparent analysis of the included studies.

A total of 18 studies were selected, providing valuable insights into the effectiveness of interventions and informing clinical practice and future research

Flowchart of the systematic review process, detailing screening, eligibility, and study inclusion with color-coded stages.

Figure 2 this diagram visualizes the PICO framework used to structure the research focus. It highlights the Population (e.g., women with early pregnancy loss), Intervention (e.g., mifepristone + misoprostol), Comparator (e.g., placebo or standard care), and Outcomes (e.g., pregnancy rates, mental health improvement). The framework is represented with distinct colors and connecting arrows to illustrate relationships between components.

Data extraction: •

Study characteristics: Year, study design, population size, geographic location, and intervention type.

Data Extraction Table 1 summarizes the characteristics and outcomes of the 18 included studies. The table includes details on study design, population size, location, interventions implemented, outcome measures, intervention details (e.g., duration, mode, and components), and quality assessment using tools like the Cochrane tool and NOS. Limitations for each study are also highlighted to provide context to the results.

Table 2 random-effects models were used to calculate pooled prevalence rates for depression, anxiety, and PTSD. This approach accounts for variability (heterogeneity) between studies. Weights were assigned to each study based on its precision (sample size and variance). Depression had the highest pooled prevalence at 39% (95% CI: 36%–42%), indicating that nearly 4 in 10 individuals experience depression following pregnancy loss. Anxiety was the second most prevalent condition at 30% (95% CI: 27%–33%).PTSD showed the lowest pooled prevalence at 17% (95% CI: 15%–19%), suggesting it is less common but still significant. 95% confidence intervals were calculated for each pooled prevalence rate to indicate the range within which the true prevalence is likely to lie.

Figure 3 the bar chart further emphasizes depression as the most prevalent condition, with distinct gaps compared to anxiety and PTSD. The narrow confidence intervals indicate precise estimates, though some variability remains across studies and subgroups. The overlap in confidence intervals between anxiety and PTSD suggests moderate consistency, though subgroup differences highlight the importance of further stratified analysis.

Table 3 depression prevalence was highest among individuals experiencing stillbirth ( 42%) compared to miscarriage ( 38%) and ectopic pregnancy ( 37%).Anxiety and PTSD rates were also slightly higher for stillbirths.

Table 4 prevalence rates were grouped by type of loss (miscarriage, stillbirth, ectopic pregnancy) and geographic region (Asia, Africa, Europe). Mean prevalence rates for each subgroup were calculated to identify trends. Depression prevalence was slightly higher in Europe ( 41%) compared to Africa ( 40%) and Asia ( 37%), potentially reflecting differences in healthcare access, cultural factors, or study designs. Anxiety and PTSD showed similar regional patterns.

Table 5 and Figure 4 bar chart compares the pooled effect sizes of telemedicine and in-person interventions, Telemedicine: Pooled effect size (SMD/OR) is 0.59 (95% CI: 0.48–0.70), suggesting moderate effectiveness. In-person : Pooled effect size (SMD/OR) is 0.85 (95% CI: 0.72–0.98), indicating higher effectiveness compared to telemedicine.

The error bars represent the 95% confidence intervals, showing that in-person interventions consistently have greater effect sizes and narrower variability.

Assess heterogeneity (I ² ): ○

Calculate the I ² statistic based on provided data or simulated values.

Conduct sensitivity analyses: ○

Exclude studies marked as having a “high risk of bias.”

Table 6 includes both textual and numerical representations of risk levels, along with detailed descriptions. Risk Levels: Low (1): Strong methodological rigor., Moderate (2): Some concerns, such as self-reported data or unblinded designs. High (3): Significant limitations, including lack of follow-up or diagnostic challenges.

Figure 5 the traffic light graph visually represents the risk of bias levels for each study. Green (Low): Studies with robust methodological quality and minimal bias concerns. Yellow (Moderate): Studies with some methodological concerns, such as single-center designs or limited follow-up. Red (High): Studies with significant issues, including lack of long-term follow-up or high potential for confounding. The majority of studies fall under moderate risk, indicating room for improvement in study designs. A few studies have high risk, emphasizing caution when interpreting their results.

Bar chart showing I ² statistics for heterogeneity: Original (78.7%), Sensitivity (76.1%), indicating persistent variability.

Funnel plot for publication bias

Figure 7 the funnel plot is a visual tool used to detect publication bias in meta-analyses by plotting the standard error of each study against its effect size. The effect sizes (dots) are symmetrically distributed around the mean effect size (blue dashed line), suggesting no strong evidence of publication bias. The gray shaded area represents the 95% confidence interval (CI) around the mean effect size. Most points fall within this range, indicating consistency. A lack of studies with low standard errors (at the top) could hint at potential under-reporting of studies with non-significant results.

Table 7 Egger’s regression test statistically evaluates the asymmetry in the funnel plot to detect publication bias. Egger’s Intercept: 7.817.817.81 — Represents the magnitude of asymmetry. A higher intercept indicates potential bias. P-value : 0.2130.2130.213 — Since this value is greater than the standard threshold (p<0.05p < 0.05p<0.05), it suggests no statistically significant evidence of publication bias. The funnel plot symmetry is supported by the Egger’s regression results, reinforcing the conclusion that publication bias is unlikely to substantially affect the included studies.

This Table 8 presents a comprehensive thematic synthesis of the 18 included studies, focusing on patient and provider experiences, barriers, and facilitators. Patient experiences highlight feedback and outcomes from interventions, while provider experiences focus on practical challenges and implementation ease. Barriers detail limitations such as study design, population diversity, and data quality. Facilitators emphasize the strengths and potential of interventions to improve psychological and clinical outcomes.

Table 9 this table presents the subgroup analyses of 18 studies, categorized by type of pregnancy loss, intervention type, region and healthcare setting, and key findings. It highlights the diversity of interventions (e.g., art therapy, telemedicine, pharmacological approaches) and their impact on clinical and psychological outcomes. The table also emphasizes the geographic and healthcare context (developed vs. developing regions) and offers insights into the specific findings for each intervention and population group.

This table 10 summarizes the meta-regression analysis examining the impact of intervention duration, delivery method, and population characteristics on psychological outcomes. It includes a diverse range of interventions (e.g., in-person therapy, telemedicine, and pharmacological approaches) and highlights key psychological outcomes such as emotional health, satisfaction, pregnancy rates, and mental health disparities.

Impact of duration on outcomes

Intervention duration: Longer interventions, like multi-session therapies and longitudinal monitoring, yielded better psychological outcomes compared to shorter interventions.

Delivery method: In-person interventions showed consistent positive results, while telemedicine and self-reported methods highlighted limitations like recall bias and disparities.

Population characteristics: Larger and multi-center studies produced more robust findings, while smaller, single-center studies faced generalizability issues.

Key outcomes: Combination therapies and immediate FET demonstrated significant psychological benefits, while disparities in mental health remain a critical area for improvement.

Linear regression results: •

Slope: Indicates the rate of change in psychological outcomes with increasing duration.

Table 11, Slope (-0.06): The slight negative slope indicates a minimal decrease in the dependent variable (e.g., psychological outcomes) as the independent variable (e.g., intervention duration) increases. However, the change is negligible and lacks practical significance. Intercept (6.86): This represents the predicted outcome value when the independent variable is zero. It serves as the baseline measure. R-squared (0.01): The model explains only 1% of the variability in the outcomes, indicating a very weak relationship between the variables. P-value (0.651): Since the p-value is greater than the threshold of 0.05, the relationship is not statistically significant. This suggests that the independent variable (e.g., duration) does not have a meaningful impact on the dependent variable. The linear regression analysis indicates no significant association between the independent and dependent variables. Other factors or more complex models may better explain the variability in outcomes.

Figure 8 this scatter plot illustrates the relationship between the duration of interventions (measured in approximate sessions or months) and psychological outcomes (numeric scores). Data points (yellow crosses) represent individual studies, while the regression line (red) indicates a trend of decreasing psychological outcome scores with increased duration. Despite slight variability in data points, the regression line suggests a potential inverse correlation between intervention duration and psychological outcomes. This trend may reflect diminishing returns on outcomes over extended durations or variations in study designs and populations.

Psychological interventions were found to significantly improve emotional well-being and reduce stress among women experiencing pregnancy loss or related challenges. Zahmatkesh et al. (2024) demonstrated the efficacy of art therapy in enhancing the quality of life and reducing grief among 60 women who had experienced recent pregnancy loss. Through structured emotional processing, participants achieved better coping mechanisms and mental health improvements compared to the placebo group. Similarly, Jensen et al. (2024) showed that meditation and mindfulness interventions reduced perceived stress in women with recurrent pregnancy loss. These findings suggest that tailored psychological therapies can address the unique emotional needs of this population, offering an effective supplement to routine care. ⁴

Cognitive-behavioral therapy (CBT) was another intervention of interest. Pettman et al. (2023) ¹⁸ explored task-sharing approaches to deliver CBT for antenatal depression in resource-limited settings. While the intervention improved adherence to prenatal care, it did not significantly reduce depression scores. This highlights the challenges of implementing psychological interventions in under-resourced settings, where additional support mechanisms may be required to achieve measurable improvements in mental health outcomes.

Clinical outcomes were primarily driven by pharmacological and ART-related interventions, with notable successes in improving pregnancy rates, tissue expulsion success, and maternal satisfaction. •

Pharmacological therapies: Mifepristone combined with misoprostol emerged as a highly effective intervention for managing early pregnancy loss. Shimels et al. (2023) ¹⁹ demonstrated that this combination significantly increased tissue expulsion success rates and improved patient satisfaction compared to misoprostol alone. Hamel et al. (2024) corroborated these findings, highlighting the cost-effectiveness of mifepristone and misoprostol in a U.S. cohort.

Low-dose aspirin also showed promising results. Naimi et al. (2021) ¹¹ found that daily administration of 81 mg aspirin significantly improved live birth rates and reduced pregnancy loss in women with prior losses. This intervention provides a simple, scalable solution to improve pregnancy outcomes, particularly in high-risk populations. ¹⁰

Abbas et al. (2023) investigated the use of home-based misoprostol for postpartum hemorrhage management in rural India. While the intervention was deemed safe, it did not significantly reduce hemorrhage rates. These findings suggest that misoprostol may require complementary strategies to achieve meaningful clinical benefits in resource-limited settings.

Subgroup analyses provided valuable insights into the differential impacts of interventions based on pregnancy loss type, geographic region, and healthcare settings. •

Type of pregnancy loss: Pharmacological interventions, particularly mifepristone and misoprostol, were most effective for managing miscarriage-related outcomes. Immediate FET protocols, on the other hand, demonstrated significant benefits for women with ectopic pregnancies, resulting in higher clinical pregnancy rates (Gao et al., 2024). ¹³

The pooled analysis showed substantial heterogeneity indicating variability across study designs, populations, and interventions. Sensitivity analyses excluding high-risk studies reduced heterogeneity slightly but remained high, reflecting inherent differences in the included studies.

High heterogeneity highlights the need for more standardized protocols in future research. While the included studies varied in their methodologies, their collective findings provide a robust evidence base for guiding clinical practice.

Visual assessment of the funnel plot and Egger’s regression test suggested no significant evidence of publication bias. However, the possibility of underreporting nonsignificant findings cannot be entirely excluded. Future meta-analyses should aim to incorporate gray literature and unpublished studies to minimize potential bias. ¹⁵

The thematic synthesis provided insights into patient and provider experiences, as well as the barriers and facilitators influencing intervention outcomes. •

Barriers: Key barriers included small sample sizes, single-center designs, and lack of long-term follow-up in many studies. Shorter et al. (2020) ¹⁶ highlighted racial disparities in mental health outcomes, particularly among Black women, emphasizing the need for culturally tailored interventions. Self-reported data, as used by Tempest et al. (2022), ²² were prone to recall bias, limiting the reliability of findings.

Meta-regression analyses revealed that intervention duration and delivery method significantly influenced outcomes. Longer intervention durations, such as multi-session therapies and longitudinal monitoring, were associated with improved psychological and clinical outcomes. Delivery methods also played a critical role; in-person interventions consistently outperformed telemedicine approaches, which faced challenges related to accessibility and equity. ^{13, 18}

This review had several limitations. High heterogeneity across studies reduced the precision of pooled estimates, and the reliance on self-reported data in some studies introduced potential biases. Additionally, the exclusion of non-English publications may have limited the generalizability of findings to non-Western settings. Addressing these limitations in future research could enhance the robustness of evidence. ²